Mobile & Marine Robotics Research Centre Challenges of ROV System Integration for Deep Water Habitat Mapping University of Limerick Mapping cold water corals at 800-1,200 m on the margins of the continental shelf off west coast Ireland Dan Toal
Overview of Talk Mobile & Marine Robotics Research Centre HEA PRTLI 3 - Deep Ocean Habitat Mapping Project High Resolution Imaging with ROV Deployed Sensors, Positional Accuracy v Resolution Trade Off and Importance of Nav/positioning Systems Integration ROV, Ship, Sonar & INS Simulator Development. VR Command Centre, & Real-time Imaging Sonar Simulator Research Cruise CE0505 June 2005
Mobile & Marine Robotics Research Centre Dan Toal, Edin (dino) Omerdic, James Riordan, Levente Molnar Postgrads: S. Nolan, E. Thurman, J Horgan, H Ahmad Main Thrust of Research Intervention Autonomous Underwater Vehicles Robot operation close to the seabed and marine structures Navigation & control Artificial intelligence, neural-networks Imaging sonars Sidescan & Multibeam. Simulators, Immersive VR Command Centre Vision in vehicle control. Robot arms for intervention. Proximal object detect sensors for operation near seabed
Tethra & Mini-Sub AUVs Wheeled robots, robot arms Marine survey equipment Reson 7125 Multi-beam Sonar Sound Velocity Profiler Tritech SS sonar Tritech Colour Camera. Navigation / positioning Ixsea Fibre Gyro INS RDI Doppler Velocity Log Microbath precision depth & altimiter sensor Obstacle avoidance 6 Tritech altimeters Thrusted Pontoon Mobile & Marine Robotics Reseaech Centre Facilities & Equipment: Robots
Test Pool and 50m Arena Pool 4.5m deep test pool 50m Olympic pool
HEA PRTLI 3 Deep Ocean Habitat Mapping Using a Remotely Operated Vehicle HEA PRTLI 3 project In collaboration with Department of Earth & Ocean Science, NUIGalway, Capital for ROV Toolskid (Multibeam + Precision Navigation) Funding delayed by 18 month Year 1 Tender, design, integrate Year 2 Shakedown cruise + Science Year 3 Deep ocean mapping Year 1 - Year 2 Tender Year 3 Design Integrate, Do or Die Cruise.
Operations Near Seabed Arc Mounds Belgica Mounds Celtic Explorer ROV Deployment
Sonar Imaging of the Seabed (Sidescan, Multibeam), Positional Accuracy v Resolution Trade Off & Importance of Navigation / Positioning
Experimental Sidescan Data Logger Geoacoustics 100/410 khz sidescan sonar image acquired with a NI DAC card interfaced to a laptop PC Custom Labview application as the data acquisition software Echo data fused with concurrently generated GPS and IMU data in XTF format. Screen shot of data gathered in Survey of Clewbay, Ireland
Sonar Imaging of the Seabed Multibeam Dynamically positioned ship (DGPS) Time to reception of echo for each beam gives 120 range / angle pairs across the track of the ship. 120 range angle pairs per ping as the vessel moves builds up the Bathymetry (shape of seabed). Assumes velocity of propegation, position and attitude of vessel/sonar are known
Sonar Imaging of the Seabed Sidescan & Multibeam Necessary for building up image 1. Sonar returns 2. Knowledge of position & orientation of sonar / vehicle 3. Measure of motion disturbance of platform 4. Refraction of beams due to Sound Velocity Profile (SVP) Errors in measurement and/or approximations all result in degradation of imagery. Fuzzy out of focus images Sonar processing / filtering result in averaging (smoothing).
Sonar Imaging of the Seabed Effects of Platform Motion Roll Heave Yaw Roll beam footprint shifts laterally Yaw beam footprint rotate Pitch Sway Surge Pitch beam footprint shifts forward or back Heave afects dimension of footprint Surge & Sway - afect position of footprint Combination of all six disturbances? Beam refraction and range / delay? Unknown geometry of bed, Instrument misalignment - patch test.
Motion Disturbance of Platform some figures on the effects Assume: 1,000m depth, depression angle 30, Fan beam - 1 along track, 120 beams equiangular beam spacing - 1 across track 1 5 Pitch offset Roll offset Yaw offset 17m along track, 1 beam width Outer beam 72m across track Outer beam 30m along track 87m, 5 beams 412m across track 150m along track
Variation in Sound Velocity Refraction
Error at 500m depth Sound Velocity
Marine Survey Ship Mounted Multibeam Dynamically positioned ship (DGPS)
Marine Survey ROV Mounted Multibeam Higher resolution imagery Lower positional accuracy, narrow swath. Wide area surveying and image mosaicking? Dynamically positioned ship (DGPS) Tether management system, depressoor weight ROV SBL/USBL transponder AUV
Position accuracy on sea surface GPS Source Uncorrected With Differential Ionosphere 0-30 meters Mostly Removed Troposphere 0-30 meters All Removed Signal Noise 0-10 meters All Removed Ephemeris 1-5 meters All Removed Clock Drift 0-1.5 m All Removed Multipath 0-1 meters Not Removed SA 0-70 meters All Removed DGPS much higher level of accuracy than GPS, RTCM corrections. Real Time Kinematic (RTK) GPS positional accuracies to ~ 10 20 cm. Roving receiver limited to within 10km radius of fixed receiver - in shore applications. Networked RTK GPS for offshore use e.g. Fugro multiple fixed receivers with satellites can extend the range of rover GPS receiver. Sub meter accuracy possible.
Nav sub systems DGPS, Ship MRU, USBL(GAPS), INS, DVL, Precision Depth Position Accuracy Subsea Motion Reference Unit GAPS USBL 1% Slant range Fibre Gyro INS + Kalman Filter Doppler Velocity Log Precision Depth Position error for sub sea: INS < 3m / hour drift bounded by 1% slant range USBL - (10 15m error at 1,000m depth) Image resolution of 0.1m?
Positional Accuracy / Resolution Trade Off. Ship mounted Sonars High accuracy Low resolution Wide swath/ field of view Regional scale multibeam bathymetry/ backscatter mapping (National Seabed Survey) Hull mounted Sonars Kongsberg-Simrad EM120, EM1002 Low resolution survey 1-10m seabed feature size typically distinguishable ROV mounted Sonars Low accuracy High resolution Narrow swath/ field of view High resolution multibeam bathymetry/backscatter mapping ROV mounted sonars. Reson 7125 Multibeam Video mosaic mapping from ROV 0.1-1m seabed feature size typically distinguishable 0.01-0.1 m seabed feature size typically distinguishable High resolution survey
Mobile & Marine Robotics Research Centre Systems Integration ROV, Ship, Sonar & INS University of Limerick
Navigation Equipment RV Celtic Explorer GPS1 GAPS USBL GPS1 GAPS Umbilical Tether ROV Wet Bathysaurus ROV PHINS Mobile GPS2 Depth sensor DVL USBL transponder Bottle Toolskid Fibre converter GPS2 MUX PHINS Wet Bottle Rs232 split DC PSU 7LCU 7125 Sonar System SVP RDI DVL Interconnect Wet Bottle Microbath Precision Depth & Altimeter
Technical challenges GAPS PHINS Interface (incompatible protocols) Send depth sensor data to GAPS GAPS Holder Depth sensor GAPS Interface GAPS PHINS
Technical challenges Depth sensor PHINS Interface (incompatible protocols) Hardware solution Depth sensor Depth (Microbath format) Onboard microcontroller Depth (PAROSCIENTIFIC format) Software solution PHINS Depth sensor Depth (Microbath format) Software on mothership Depth (PAROSCIENTIFIC format) PHINS
Technical challenges DVL (Real-time monitoring & Real-time change of mode of operation) Send corresponding configuration scripts during different stages of mission DVL Real-time monitoring Configuration scripts: Water track only Bottom track only Water & Bottom track (combined) Software on mothership PHINS (Real-time monitoring & Real-time configuration) Electrical interface (communication parameters, protocols etc.) Mechanical interface (lever arms of sensors, geometric configuration etc.) Real-time monitoring Real-time configuration Software on mothership PHINS
Preparation Interconnection diagrams (two variants) Quick Reference Book (collection of key information from different manuals in compact form) MATLAB software to determine lever arms and make easy PHINS configuration LabView software for GAPS-PHINS-DEPTH interface DVL test trials in UL diving pool in order to test sign of DVL outputs
Interconnection Design
Problems during mobilisation / shakedown Leak in the interconnection bottle Cause: faulty connector Responsibility: OceanTools (manufacturer) Solution: Running repare of faulty connector Wrong operation of PHINS (instrument rotated by 180 inside enclosure & bugs in firmware) Cause: error in technical drawing of enclosure Responsibility: system integrator (RESON) Solution: Rotation of PHINS and firmware update Multibeam (7125) Integration. Day before embarking advised to swap sonar. Cause: New product, DOA Responsibility: (Reson) Solution: Panic & swap to mature 8125 GAPS data were not always accepted by PHINS Cause: wrong setting in GAPS software Responsibility: IxSea engineer & confusing GAPS software Solution: Set parameter D to correct value D = z SL +z BOS +z DK +z H = 12.545 m Distance Value Unit Type z SL 1.363 m Variable z BOS 7.262 m Fixed z DK 3.000 m Variable z H 0.920 m Fixed Level BOS Drop Keel Holder Sea Level CG z SL z BOS z DK z H GAPS
Results Successful integration GAPS-PHINS-DEPTH for the first time! Accurate determination of lever arms (all sensors were accepted by PHINS). High accuracy of navigation data. Succesful DVL real-time reconfiguration. Systematic approach for system integration. Discovery of errors in Bloom report.
Identified weak points Lack of leak detectors Inability to remotely switch on/off individual components Full separation between control and navigation Difficult navigation with existing displays in severe conditions
Simulation Solutions Virtual Laboratory (A talk for another day) Modelling of pools, marine environment, vehicles (ship AUVs and ROVs), navigation & payload sensors with signal level compatibility. Hardware in the loop testing and progressive porting of developed code to the realtime target. Shift the load of integration and debugging to offline simulation rather than during mobilisation with daily ship cost. VR Command Centre for Offshore Ops (Augmented Reality) Video Clip Dr Edin Omerdic Real-Time High Resolution Imaging Sonar Simulator Dr James Riordan
VR Command Centre for Offshore Ops (Augmented Reality) Lunar Module calling Heuston Control, Come in Heuston Roger niner, this is Heuston Features Library of Vehicles (Ships, ROVs, AUVs) Full dynamic models with signal-level compatibility vr models integrated with real-time sensor input - augmented reality during ops. Configuration support tools for new vehicles. Fault accommodation capabilities Background seabed map import facility (INSS) Models of waves, 3 D current models and more.
VR Command Centre for Offshore Ops (Augmented Reality)
Real-Time High Resolution Imaging Sonar Simulator As discussed Imaging the Deep Many issues affecting the quality of images, maps, 3D generated geometry model Without draining the ocean it s impossible to qualify?
Seafloor Terrain Synthesis Model Multi-Fractal Surfaces Hybrid Model Fractal Dimension varying with altitude Combination of: Perlin noise synthesis Fourier synthesis Wavelets Permits local control over terrain features. Multiple geologic features can be incorporated in a single terrain environment. Peaks and outcrops rough, valleys smooth erosion and deposition Low lying regions consist of sediment ripple beds influence of currents
Acoustic Propagation Model Path of acoustic pulse determined by ray theory solution to Helmholtz Eq. Seafloor Scattering Model Output of Jacksons backscattering model for a typical seafloor configuration
View Dependant Refinement of Fractal Model of Seabed. Full resolution terrain mesh viewed with adaptive refinement disabled Multi-resolution representation of the terrain mesh, adaptive refinement enabled.
View Dependant Refinement Full high-resolution mesh. The sonar position and fan beam footprint are represented by the superimposed white sphere and blue wedge respectively. Multi-resolution abstraction of the seafloor mesh. The geometric detail within the fan beam footprint is preserved while non-contributing facets are decimated to the maximum possible.
Incremental View Dependant Refinement Real-time view dependant Level Of Detail control of terrain mesh Detail information reserved for recovering M i+1 from M i. The structure queried efficiently online during simulation to obtain adaptive meshes on-the-fly by reconstructing the vertex dependency relations. Multiple sonar footprints can be accommodated.
Results Multi Fractal SS Imagery
Results - Image Generation Performance Based on Simulator Experimentation 4-5 Hz ping rate while imaging seafloor terrain ~ 2 million facets. 40-fold increase in simulation throughput. Achieved principally by use of video rendering techniques within sonar simulation process. No degradation on integrity of imagery. Improved speed and quick results Simulation a more useful tool Allows user interaction on the fly as simulation progresses. Performance capability extended to survey scale maps.
Integration of Simulators Virtual Laboratory vr Command Centre for Offshore Ops Real-time High Resolution Imaging Simulator Work Underway
HEA PRTLI 3 MSR3.2 R.V. Celtic Explorer Research Cruise CE0505 June 2005 First Irish-led deep water ROV habitat survey technical and scientific partnership NUIG/UL ROV multibeam and video surveys high resolution mapping and nondestructive observations of habitat/fauna Use INSS data to identify survey targets and plan surveys essential data for cost effective survey Bathysaurus ROV
Margaret Wilson, February 2006
ROV & INSS multibeam